What disaster recovery really costs
DR cost comes down to one question: how much of your production setup has to be running at the DR site before a disaster happens? Answer that for each application and most of the budget follows.
Where the money goes
- Space and power at the DR site. Your own second data centre, a colocation rack, or a cloud account.
- Compute that runs every day. Servers kept powered on for databases, directory services and any warm or active tiers.
- Storage. A full copy of protected data, plus snapshots and backup versions. Plan for more than one copy's worth of capacity.
- Network links. Replication bandwidth between sites, ideally from two different providers.
- Software licences. Databases, hypervisors, backup and replication tools. Some vendors charge in full for a standby copy and some do not, so read the contract.
- People and drills. Time to write and maintain runbooks and to run tests at least once or twice a year.
Rough planning ranges
For a first conversation it helps to express DR running cost as a share of what the production infrastructure costs. These are the ranges we use to compare options before detailed sizing.
| Pattern | DR running cost vs production | Why |
|---|---|---|
| Backup & restore | About 10–20% | Mostly storage and backup software. Almost no compute runs at DR. |
| Pilot light | About 20–35% | Replicated storage plus a small set of always-on core servers. |
| Warm standby | About 40–60% | The whole stack runs, at reduced size. |
| Active-passive | About 80–100% | A full-size copy that mostly sits idle. |
| Active-active | 100% or more | Two full sites, plus the cost of changing the application. |
These are planning ranges, not a quote. Use them to compare options early, and replace them with real sizing before a budget is approved.
A worked example
Say production has 40 virtual machines and 30 TB of data. After talking to the business owners, 8 of those VMs are critical (tier 1), 20 are important (tier 2), and 12 can wait a day or two (tier 3).
Option A: protect everything the same way
Build a full active-passive copy. The DR site needs hardware for all 40 VMs, running all the time, plus a full replicated copy of the 30 TB.
Option B: protect by tier
- Tier 1 (8 VMs) on warm standby at about half size.
- Tier 2 (20 VMs) on pilot light: databases running small, application servers off.
- Tier 3 (12 VMs) on backup and restore.
Day to day, Option B runs the equivalent of roughly 7 VMs at the DR site instead of 40. Tier 1 still recovers just as fast as it would under Option A.
There is a catch, and it depends on where the DR site is:
- On-premises or colocation: you still need hardware for tiers 1 and 2 (28 VMs) ready to start, because you cannot buy servers on the day of a disaster. The saving is in power, licences and the tier 3 capacity, not in all the hardware.
- Cloud DR: you pay only for the small always-on footprint and the storage. Full capacity is created on drill day or disaster day. This is where the tiered approach saves the most.
Costs people forget
- Getting data back. After a cloud failover, moving data back to your own data centre can carry large egress charges. Plan failback before you need it.
- The first copy. Seeding tens of terabytes over a WAN link can take weeks. It is often faster to ship encrypted disks for the first copy.
- Drill time. A full failover test takes several people most of a day, plus preparation. Budget for it every year.
- Keeping DR current. Patching DR templates, renewing certificates and updating firewall rules at the DR site is ongoing work.
- Licence clauses. Running a standby database or hypervisor may need its own licence, depending on the vendor and how the standby is used.
Not sure where your DR stands?
Send us your application list and a short note on how backups work today. We will come back with a plain gap review: which systems are exposed, what a realistic recovery time looks like, and what it would take to close the gap.